TORFL fails one subtest at a time, and lets you repair it for two years

The regulation lists them with their CEFR equivalents attached, which is more than most bodies do: элементарный (ТЭУ/A1), базовый (ТБУ/A2), первый (ТРКИ-I/B1), второй (ТРКИ-II/B2), третий (ТРКИ-III/C1), четвертый (ТРКИ-IV/C2).

ТРКИ состоит из 5 частей/субтестов: «Чтение», «Письмо», «Лексика. Грамматика», «Аудирование», «Говорение». One sentence, no qualifications, applying to all six levels.

Two of the levels carry a stated purpose. ТРКИ-I is the level required for admission to Russian universities at bachelor's level, and ТРКИ-II for master's programmes. The Institute states both plainly on its own page, so a candidate does not have to guess which certificate the admissions office wants.

What varies with level is time. The published duration runs 3 hours 30 minutes at ТЭУ, 4 hours at ТБУ, 4 hours 45 at ТРКИ-I, 5 hours at ТРКИ-II, and 5 hours 30 at both ТРКИ-III and ТРКИ-IV.

This comparison is ours. Nearly every laddered exam documented on this site changes its own shape as it climbs. The DALF collapses four épreuves into two at C2. Poland folds grammar into reading below B1 and gives it its own module from B1. Italy's CELI fuses skills at A1. Above B1 the Goethe modules stop meeting at all. TORFL keeps one architecture from A1 to C2 and moves only the difficulty and the clock. Whether that is a virtue depends on whether you think a C2 speaker and an A1 speaker should be asked the same kinds of question.

Section 10 of the regulation states the threshold: ТРКИ считается успешно пройденным, если Тестируемый набрал не менее 66% стоимости теста. При этом по одному из пяти субтестов допустим результат 60% при условии выполнения всех остальных четырех субтестов на удовлетворительном уровне (66 %).

So: 66%, and one of the five may come in at 60% provided the other four clear 66%. The second sentence is what fixes the meaning of the first — a 66% requirement that could be satisfied on aggregate would make "the other four at a satisfactory level" meaningless. The threshold is per subtest.

This is ours, and it is the point of this page. A sectional floor is normally absolute. The JLPT says a candidate below the sectional pass mark in one section has failed no matter how high the total score. France calls a mark below 5 out of 25 éliminatoire — the word does the work. Both are cliffs: one step over the edge and the whole result goes.

TORFL draws the same kind of line and then publishes a bounded tolerance beside it. Not a compensation rule — 60% in one subtest is not paid for by 90% in another, and the other four still have to reach 66% on their own. It is narrower and stranger than compensation: a single, one-off, six-point margin, available once across five subtests, granted to nobody who needs it twice.

That reads to us like an examining body acknowledging its own measurement error rather than softening its standard. Six points on one paper out of five is roughly the width of a marking disagreement. A cliff pretends the instrument is exact; this one says, in the regulation, that it is exact to about six points.

Section 6 is short and it changes what sitting this exam actually costs you. Повторное прохождение ТРКИ (пересдача) на условиях частичной оплаты предоставляется Тестируемым, если не сдан один субтест. Баллы по остальным субтестам засчитываются по результатам предыдущего прохождения ТРКИ. And then: Результаты успешно сданных субтестов действительны в течение двух лет.

Fail one subtest and you resit that subtest. The other four are carried over from the previous sitting, and they stay good for two years. The Institute prices it separately: a пересдача одного субтеста is 50% of the cost of the whole test.

This comparison is ours. Put it against the three answers already documented here and it is a fourth. The JLPT and the DELF are unitary in both directions: one certificate, and a failure anywhere sends you back to sit all of it. The Dutch NT2 and Finland's YKI are modular in both directions: the parts are free-standing, so a failed part was never attached to anything else. Germany's Goethe modules above B1 are the same. TORFL is unitary in the certificate and modular in the repair — you are awarded one document at one level, and yet the thing you failed is the only thing you have to do again.

The practical arithmetic is worth doing before you book. Fail one subtest at ТРКИ-II and the repair costs you 50% of 6,500₽ and one appointment. Fail one subtest on an exam that does not work this way and the repair costs a full fee and a full day — five hours forty at ТРКИ-II, four of them redoing work you had already passed.

One caution, because the regulation does not say and this site does not guess: it addresses the case where one subtest is failed. What happens when two are is not stated on the pages read here. Ask the centre before assuming the mechanism scales.

Section 4.12.2 is unusually specific about the tape. Аудиозапись субтеста «Говорение» осуществляется на протяжении всего субтеста. Аудиозапись каждого Тестируемого производится поточным способом, разрыв аудиоматериалов на отдельные фрагменты (реплики) и последующий их монтаж не допускаются.

Continuous recording of the whole speaking subtest, captured as one stream, and splitting the audio into fragments and editing them afterwards is expressly forbidden. The performance leaves a durable, tamper-resistant artefact.

This is ours, and it is the reason this section exists, because it tests something this site published earlier today. Cambridge English states that its Speaking tests cannot be re-checked, and gives the reason in its own words: the tests are not recorded, so there is nothing for a third examiner to re-read. We drew a general mechanism out of that and out of Norway's appeal rules — that appealability tracks whether a performance left an artefact.

TORFL is the case that tests it, and it comes out half right. The artefact exists here, and is protected by regulation. But the candidate's remedy is still not a re-reading. Section 7 gives a конфликтная комиссия that considers a complaint within five working days of the test, and section 6 gives the resit. Nowhere on the pages read here is the recording offered to the candidate as grounds for having the mark re-examined.

So the refined version, and we would rather state the refinement than leave our own earlier claim standing unqualified: an artefact is necessary for a re-read, and it is not sufficient. The tape here is kept for the Commission — identity, integrity, procedure — rather than for the person who was recorded. Cambridge has nothing to re-read and offers nothing. Russia has something to re-read and offers a re-performance instead.

Section 9 of the regulation is a timetable, and its third column is one this site has not seen an examining body print anywhere else. Alongside the written part and the speaking subtest, it states the time a member of the Commission spends assessing each candidate.

ТЭУ/A1 — 0.5 academic hours. ТБУ/A2 — 0.5. ТРКИ-I/B1 — 1.4. ТРКИ-II/B2 — 1.4. ТРКИ-III/C1 — 1.8. ТРКИ-IV/C2 — 1.5.

Read down that column and the exam's cost structure is visible without anyone explaining it. Assessment labour almost triples between A2 and B1 — 0.5 to 1.4 — and the published fee steps up in the same region. The examiner's hour is the expensive input, and the regulation says so with a number instead of leaving it to be inferred.

This comparison is ours. Finland's OPH reaches the same fact from the other end: it prices the YKI subtests differently from each other — listening €43, reading €43, writing €70, speaking €84 — so assessing a productive skill costs roughly twice assessing a receptive one, stated as money. Russia states it as time. Two bodies, the same underlying economics, and the only two we have read that admit publicly that marking a person speaking is not the same job as marking a multiple-choice paper.

It also quietly explains why a resit of one subtest is priced at a flat half of the whole test rather than in proportion to the part. The Institute is not pricing the paper; it is pricing an appointment and an examiner.

The Institute publishes its own prices, per level, in figures — which puts it in a minority on this site. ТЭУ 5,500₽. ТБУ 6,000₽. ТРКИ-I 6,000₽. ТРКИ-II 6,500₽. ТРКИ-III 6,500₽. ТРКИ-IV 6,500₽. A duplicate certificate is 900₽, issued where the original is lost or was made out wrongly through the candidate's own fault.

Six levels and three prices: the ladder is banded, not tariffed per hour. ТБУ and ТРКИ-I cost the same despite 45 minutes between them, while ТЭУ sits alone at the bottom.

And the figures come with their own limit attached: Указанная стоимость распространяется только на тестирование, проходящее в Институте Пушкина. В зарубежных центрах стоимость тестирования определяется соглашением о сотрудничестве. A price that is published, real, and explicitly local — so the comparison table on this site records it as published per centre rather than as a national fee.

One last detail, because it prevents a misreading. Each level's paper has a different maximum score, and the maxima are not in level order: ТЭУ 530, ТБУ 730, ТРКИ-I 675, ТРКИ-II 660, ТРКИ-III 650, ТРКИ-IV 685. The highest-scoring paper in the entire system is A2, and C2's is 45 points below it.

That is not a difficulty statement, and reading it as one is the mistake. The pass is a percentage, so the raw totals are an artefact of how many tasks each paper happens to carry. The JLPT sets a related trap from the other direction, documented elsewhere on this site: its papers are all scored out of 180 while its pass marks move by level — 100 for N1, 90 for N2 and 95 for N3 — so there the constant number is the maximum and the moving one is the threshold. TORFL is the mirror image: the threshold is a constant 66% and the maximum moves. What the task counts do show is a real curve: Лексика. Грамматика runs 70 tasks at A1, 110 at A2, 165 at B1, 150 at B2, then 100 at both C1 and C2. Explicit grammar is tested hardest at B1 and asked about less and less after it.